IEEE Power & Energy Magazine - May/June 2017 - 33

1.08
1.06

Voltage Limit

1.04
1.02
1

Voltage HC

HC Probability (%)

Maximum Network Voltage (p.u.)

1.1

10
20
30
40
50
Total Nominal Power of Installed DG (MW)
(a)

12
10
8
6
4
2
0
0

150

(a)
Loading HC

100

Loading Limit

50
10
20
30
40
Total Nominal Power of Installed DG (MW)
(b)

Single Generator Installed in the Most Remote Bus
DG Uniformly Distributed on the Network
Single Generator Installed in Proximity of the
Primary Substation
Random Position of DG Units

HC Probability (%)

Network Loading (%)

Total Nominal Power of Installed DG (MW)

12
10
8
6
4
2
0
0

Total Nominal Power of Installed DG (MW)
(b)

figure 2. Simulation results of numerous allocations of DG
compared to network (a) voltage and (b) current limits.

figure 3. HC probabilistic distributions obtained using
the combined Monte Carlo- and OPF-based approaches
considering (a) voltage and (b) current limits.

In general, the performance of distributed solutions highly
depends on the position of the DRES units. The performance
of centralized/supervised technologies, however, is almost
the same in all of the scenarios considered.
Another common factor among the best performing solutions is field measurements of a few parameters at well-selected
points in the network. These studies show that approximately
80% of the medium-voltage (MV) feeders present only one
critical node affected by maximum acceptable voltage constraints. This means that it is not necessary to deploy an enormous number of sensors to obtain the best results for increasing
HC, but it is important to select the right places to install them.
Identifying a node for measuring normal conditions is much
easier than examining different scenarios of emergency supply
for forced outages or alternative supply under system maintenance conditions. But even for normal conditions, the critical
node might change when additional DG units are installed. In
the analyzed low-voltage (LV) grids, 60% of the feeders present
more than one critical node. Since the number of required sensors is highly dependent on the connection point of the DRESs
along the feeders, this aspect must be studied case by case.

We analyzed two complete distribution grids of different
distribution system operators (DSOs) and identified voltage
violations as the most important problem when increasing the
number of DG units. For the first DSO, 90% of the feeders
were voltage constrained while only 10% were loading constrained (Figure 4). The second DSO presented 77% of the
feeders as voltage constrained, with a homogeneous distribution of DG until current constraints occur.
The horizontal axis of Figure 4 represents the percentage loading for the feeders. Most of the feeders are voltage
constrained (blue bars) and present a loading that is well
below the 100% loading limit. The black curve (cumulative
distribution function) shows that for half of the feeders, the
normal loading is below 37%. This means there is a large
reserve (more than 63%) of thermal loading capability.
This loading reserve is obtained at a homogenous DG penetration without applying any SG solution. In general, rural
feeders are voltage constrained, while urban feeders are loading constrained. The global results, however, depend on the
type of feeders. For the second DSO, for instance, the loading
reserve is about 40% for half of the feeders. From this analysis,

may/june 2017

ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - May/June 2017